A Stability Assessment Framework for Process Discovery Techniques
An extensive amount of work has addressed the evaluation of process discovery techniques and the process models they discover based on concepts like fitness, precision, generalization and simplicity. In this paper, we claim that stability could be considered as an important supplementary evaluation dimension for process discovery next to accuracy and comprehensibility, with ties to the generalization concept. As such, our core contribution is a new framework to measure stability of process discovery techniques. In this paper, the design choices of the different components of the framework are explained. Furthermore, using an experimental evaluation involving both artificial and real-life event logs, the appropriateness and relevance of the stability assessment framework is demonstrated.
KeywordsStability Process discovery Conformance checking Validity Log perturbation
- 2.Bolt, A., de Leoni, M., van der Aalst, W.M.P.: Scientific workflows for process mining: building blocks, scenarios, and implementation. Int. J. Softw. Tools Technol. Transf. 1–22 (2015). doi: 10.1007/s10009-015-0399-5
- 10.van Dongen, B.F., Mendling, J., van der Aalst, W.M.P.: Structural patterns for soundness of business process models. In: 10th IEEE International Enterprise Distributed Object Computing Conference, pp. 116–128 (2006)Google Scholar
- 18.Maruster, L.: A Machine Learning Approach To Understand Business Processes. Eindhoven University of Technology, Eindhoven (2003)Google Scholar
- 22.Vanden Broucke, S.K.L.M.: Artificial negative events and other techniques. Ph.D. thesis, KU Leuven (2014)Google Scholar
- 26.Weijters, A.J.M.M., van der Aalst, W.: Rediscovering workflow models from event-based data using little thumb. Integr. Comput. Eng. 10, 151–162 (2003)Google Scholar